What is Data Poisoning? | AI Jargon Buster | Monard X
← Back to Tools
AI Policy and Regulation

What is Data Poisoning?

Data poisoning is a security attack where someone intentionally introduces bad, misleading, or malicious information into the data used to teach an AI model. Because AI systems learn by identifying patterns in the information they are fed, an attacker can use this method to trick the system into learning incorrect rules or behaving in a specific, harmful way. It is essentially a form of sabotage that compromises the reliability and integrity of a system from the inside, often by exploiting the trust an AI places in its training sources.

Why this matters to you

It represents a critical security risk for any organization that relies on public, crowdsourced, or third-party data to build its custom AI tools. If your data sources are compromised, the resulting AI may produce inaccurate results, bypass security filters, or expose sensitive information, which can lead to significant financial or reputational damage for your business.

How you might hear this

The security team implemented strict data verification protocols to prevent data poisoning during the model training phase.

AI Jargon Buster

Search any AI term, explained in plain English.

Type a term below and search. You will be taken straight to the tool.

Career Corner Beta